Tuesday, October 08, 2013

Management of acute pancreatitis: can we be evidence based?

Evolving concepts and controversies are addressed in a recent review published in the Cleveland Clinic Journal of Medicine. The article is available as free full text. A few take home points will be mentioned here.

The optimal timing of CT scanning is based on the fact that the yield for detecting necrosis is relatively poor until 72 hours or so after presentation. Earlier scanning may be indicated in certain clinical circumstances suggestive of complications. Scanning at the time of presentation is justified if the diagnosis is unclear. That is to say that CT scanning provides just one of the three diagnostic criteria, the other two being clinical and laboratory. If those two are present CT is not necessary to make the diagnosis.

There has been a gradual shift in thinking about severity assessment. While prognostic assessment may be helpful in anticipating complications, advising patients and assembling resources it may not be as useful in an algorithmic approach to management as once thought. For example, a clinical determination of severe pancreatitis defines the patient as having pancreatic necrosis. In the traditional view that would be an indication for antibiotic therapy. But that thinking is not supported by evidence and no longer holds sway. In addition, the role of severity assessment as a target for fluid resuscitation is unclear. Traditional thinking held that assessment of pancreatitis as severe targeted patients for more aggressive initial fluid administration. A study form 2011 by Warndorf and colleagues, however, found the opposite: it was the patients with milder (so called “interstitial”) pancreatitis who benefited most, suggesting that the window of opportunity is lost once pancreatitis progresses to severe.

There are many clinical tools available for severity assessment. Traditional tools (e.g. Ranson) are falling out of favor because they require observation over time and cannot classify patients within the first 24 hours. Some more recently validated methods, particularly the SIRS determination, are simpler to use and enable severity assessment on the front end.

Current concepts in fluid resuscitation have become more nuanced. The traditional view of “more is better” was based mainly on theory, animal data and low level human studies. Despite those limitations it held sway until challenged a couple of years ago by a study I cited here. According to that study fluids in excess of 4.1L in the first 24 hours were associated with harm. The thinking is shifting from “the more the better” to “shoot for the optimum.” Under resuscitation may lead to pancreatic ischemia at the microcirculatory level triggering necrosis, mediator release, SIRS, vascular collapse and distant organ failure. Too much fluid on the other hand may lead to organ congestion and compartment syndrome.

In addition, fluid resuscitation should be front loaded. The Warndorf paper reported that patients given greater than one third the 72 hour total volume in the first 24 hours had better outcomes. Of note, that group also had lower total 72 hour volumes.

As to the actual amount to give, the CCJM review says:

Optimum resuscitation is controlled fluid expansion averaging 5 to 10 mL/kg per hour, with 2,500 to 4,000 mL given in the first 24 hours.

The way to accomplish this in an average sized individual is to front load in the first several hours. Such a protocol was described in a 2011 paper by Wu and colleagues. It is based on an aggressive initial bolus followed by high volume maintenance fluid rates for the first few hours but then calls for a sharp reduction in IV rate if the BUN comes down a bit at either of two lab checkpoints starting around 8 hours following presentation. Patients treated according to that protocol received less total fluid than those who were not and the total amount received came fairly close to the upper range recommended in the CCJM review. Outcomes were not improved with use of the protocol but the patient numbers were small. What was found was that patients treated with lactated Ringer's did better than those treated with saline whether on or off protocol. Again the numbers were small and the outcomes were soft: SIRS and CRP.

Though based only on physiologic rationale (lower microcirculatory pH in the pancreas is believed to activate enzymes and promote inflammation) and the one small study with soft endpoints the CCJM review recommends LR over saline.

Additional discussions in the review focus on nutrition (enteral is better than parenteral, NG is as good as jejunal, and enteral nutrition reduces infection), compartment syndrome (an emerging concern in pancreatitis) and the management of walled off necrosis and fluid collections.


No comments: